Direct Perception of Three-Dimensional Motion from Patterns of Visual Motion

Abstract

Measurements of retinal motion along a set of predetermined orientations on the retina of a moving system give rise to global patterns. Because the form and location of these patterns depend purely on three-dimensional (3D) motion, the effects of 3D motion and scene structure on image motion can be globally separated. The patterns are founded on easily derivable image measurements that depend only on the sign of image motion and do not require information about optical flow. The computational theory presented here explains how the self-motion of a system can be estimated by locating these patterns.